On Wed, 20 Sep 2017 11:12:04 -0700 Marc MERLIN <marc_...@merlins.org> said:
> On Sat, Sep 16, 2017 at 11:21:58AM +0900, Carsten Haitzler wrote: > > > Any ideas what this could be? > > > This is totally destroying my battery life. > > > > that smells like the animator. intel graphics? for intel we open the dri > > device > > Correct, IGP, kernel 4.12. After my system crashed, I went back to 4.11 > and the 100% CPU loop is gone for now. > > > and ask the device driver to send us vsync/vblank interrupts for animation. > > this is what actually keeps our framerate perfectly synced with screen > > refresh for smoothness. if this is spinning in a tight loop reading from > > that device ... then something seems wrong with the device driver. it > > should only be producing events every "frame refresh" and even then we only > > turn this on when we need it (when we need to animate or get a frame tick). > > if no frame tick is on we don't request a schedule of the vsync event. if > > we're getting them from the device again and again even if we haven't > > asked... see above. that's what it smells like. did this start recently > > after a kernel upgrade or change? perhaps a libdrm change (we actually just > > find libdrm and dlopen/dlsym that and use it). this code was written many > > years ago (3? 4? 5? 6? i forgot...) and has worked solidly since then... > > It did happen after I went to 4.12, so you may be right there. I'm back > to 4.11 for now to see how that goes and then I'll try 4.12 again to see > if it comes back. this is seemingly specific to your kernel config and/or setup and/or revision of gpu. intel gpu's on my various laptops with 4.12 and 4.13 don't do this... :/ > > maybe strace with timestamps to double-check... as you should find those > > reads from fd 28 about 1 frame (1/60th or so of a second depending on > > refresh) apart. note if you have 2 screens you may get an event per screen > > as well - so more often and maybe at different times. > > Ah, timestamps, sorry. I'll get that next time. > > On Sat, Sep 16, 2017 at 11:14:50AM +0900, Carsten Haitzler wrote: > > On Fri, 15 Sep 2017 13:13:06 +0000 Stephen Houston <smhousto...@gmail.com> > > said: > > > > > If you are using any modules that have threads such as cpufreq or > > > temperature, there was a bug floating around where the threads weren't > > > sleeping and causing something similar. Try removing those and see if it > > > resolves. > > > > that was in git master, not e21. > > My only widgets are volume control and panel brightness. They seem ok. > > So let's say for now that it may be the kernel. I'll report back when I > have more info, thanks. you seem to have hit an interesting issue... :) -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- Carsten Haitzler - ras...@rasterman.com ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ enlightenment-users mailing list enlightenment-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-users
